Data-Driven Spectrum Trading with Secondary Users' Differential Privacy Preservation
Jingyi Wang,Xinyue Zhang,Qixun Zhang,Ming Li,Yuanxiong Guo,Zhiyong Feng,Miao Pan
DOI: https://doi.org/10.1109/tdsc.2019.2892447
2021-01-01
IEEE Transactions on Dependable and Secure Computing
Abstract:Spectrum trading benefits both secondary users (SUs) and primary users (PUs), while it poses great challenges to maximize PUs' revenue, since SUs' demands are uncertain and individual SU's traffic portfolio contains private information. In this paper, we propose a data-driven spectrum trading scheme which maximizes PUs' revenue and preserves SUs' demand differential privacy. Briefly, we introduce a novel network architecture consisting of the primary service provider (PSP), the secondary service provider (SSP) and the secondary traffic estimator and database (STED). Under the proposed architecture, PSP aggregates available spectrum from PUs, and sells the spectrum to SSP at fixed wholesale price, directly to SUs at spot price, or both. The PSP has to accurately estimate SUs' demands. To estimate SUs' demand, the STED exploits data-driven approach to choose sampled SUs to construct the reference distribution of SUs' demands, and utilizes reference distribution to estimate the demand distribution of all SUs. Moreover, the STED adds noises to preserve the demand differential privacy of sampled SUs before it answers the demand estimation queries from the PSP. With the estimated SUs' demand, we formulate the revenue maximization problem into a risk-averse optimization, develop feasible solutions, and verify its effectiveness through both theoretical proof and simulations.